R UTHERFORD S CHOOL P HYSICS A DDITIONAL P ROBLEMS

A COLLECTION OF PROBLEMS ABOUT ARCHES , BRIDGES , HANG ING CHAINS , CHAIN FOUNTAINS , . . .
Introduction
These questions start with simple AS level statics and escalate slowly to A2 and beyond. The skills that
are practised here are those that you would have to use as second nature if you go on to do problems in
astrophysics, high energy physics, cosmology, . . .
There are still remarkable and unsolved problems in classical mechanics.
Have you seen Steve Moulds chain fountain at http://stevemould.com/
siphoning-beads/? Watch the chain climb out of the pot, apparently defying
momentum conservation; see also figure 1. This collection of problems eventually explains phenomenon, which is really a research problem1 . On the way, we
deal with chain statics, dynamics, arches, and bridges, all problems with common mathematical and physical elements.
We urge you to try all these problems. Remember, only by doing the problems
yourself will you take active control of your own physics. If you get really stuck,
you can email the RSP Project for a pdf of hints.

The nature of tension in a chain or rope

Why, if there are forces in a rope (being pulled by forces f and f at its ends),
do the parts of the rope not accelerate? The internal system of forces are actually
tensions. At each point they act equally but oppositely to put the body under
tension; see the boxed detail in figure 2 where a rope is shown in tension from
the forces acting at each of its ends.
Consider the central slice dx of the rope. If T ( x) = T ( x + dx) there is no net force
on the slice and it remains unaccelerated. For slices at the ends of the rope there
is a tension acting on the internal side of the slices, but not at the external sides.
On these slices, external forces f are needed so there is a force balance for each
slice and they dont accelerate. For this to hold, clearly the magnitudes f = T.
The f s are forces, but it is they that create the tensions in the rope/chain.
dx

-f

T
x=0

T(x)

T(x+dx)
x x+dx
T

Figure 1: Steve Mould

holding a tub of chain
aloft, demonstrating a
fountain effect as the
chain falls to the ground.

x=L
T

Figure 2: Tensions in a rope being pulled at its ends.

Tensions when hanging under gravity

A chain hanging from one end doesnt have tensions that are constant along its length. The difference now
between T ( x) and T ( x + dx) acting on a slice is needed to support the weight ( dx) g. Here is the mass
per unit length of the rope and hence ( dx) is the slices mass. Multiplying by g gives the slices weight.
Draw a diagram similar to figure 2, but vertical, to get a feeling for the forces in a chain hanging from one
end, where x = 0.
1 See JS Biggins and M Warner, Proceedings of the Royal Society A, 2014.

http://dx.doi.org/10.1098/rspa.2013.0689

Exercise 1: A chain of total length L and mass per unit length is hanging vertically from one end where a
supporting force f is applied to stop the chain falling. By considering the mass of chain below x, give an
expression for the tension T ( x), where x is measured from x = 0 at the upper end of the chain. [MW]
An alternative method involves differentials do this if you have met Taylor expansions2 and differential equations. The difference in tensions supporting the weight can be expressed in the form:
T ( x) T ( x + dx) = g dx.
From this relation one could derive a simple differential equation for T ( x) with a simple solution

The nature of chains, ropes, masonry, arches

Chains, ropes, threads etc. share the property that their internal forces (tensions) must be along their
lengths tangential to their curves, to be mathematically precise. They have negligible resistance to bend,
at least at long enough length scales, so that if there were a sideways component to the forces within them,
they would deflect until forces were again tangential.
Masonry (stone, brick, . . .) has a similar character, but in reverse. It is strong in compression, but weak
in extension or shear (deflection). The pairs of forces T seen in a rope are now reversed to be in compression.
A chain supported at both ends forms a classical shape called a catenary 10
(first studied by Hooke in the 1670s, and described mathematically by Leiby/a
niz, Huygens and Johann Bernoulli in 1691); see fig. 3. The chains own
weight is supported by a system of tensions, varying along the chain, that
are tangential. An masonry arch has to support its own weight by a system of tangential compressions. Take the shape of a hanging chain: reverse
gravity to g (stand on your head!), and you have the ideal arch pointing upwards and the tensions reversed from those of the downward pointing chain). This technique of determining arches goes back to the Middle
Ages; wax on a hanging thread solidifies it, allowing it to be turned over as
an arch with tangential compressions.

s
The shape of a hanging chain 1
Exercise 2: Consider a chain supported at its ends and hanging under gravity. Let the mass per unit length be and measure lengths s along the chain
from its mid-point (the lowest point, since we consider the case where the
end supports are at the same height). Let T (s) be the tension at s along the
chain. Show that the shape it adopts, the catenary, has:
tan (s) =

s
a

-3

T(s)
0

x/a

Figure 3: A catenary formed

by a chain supported at its two
ends. Ex. 7 shows it has the form
y/a = cosh( x/a). The angle between the tangent and the x axis
is . Tensions T (s) are shown, in
particular at s = 0.

where (s) is the angle that the tangent to the chain at a distance s makes
to the horizontal, and a is a length you should define in terms of T (0),
and g.
This form is known as the intrinsic form of the curve (as against, for
instance, the Cartesian form).
Hint: Consider the static equilibrium of a section of chain extending from s = 0 to the point s. The tension
T (0) at s = 0 is horizontal, with T (s) at s having an angle (s) to the horizontal. The horizontally and
vertically resolved tension and weight forces must balance. The mass between the lowest point and s is s,
the weight of which of course acts vertically. [MW & PS.]
2 T(x

+ dx ) = T ( x ) +

dT
dx

dx + . . . ; see the relevant Maths concepts pages.

3
3

The dynamics of chains

When a chain is pulled into motion by a force applied at one end, both momentum and kinetic energy are
created. It is deeply worrying that these turn out to be at variance with each other, as the problem below
illustrates. First some simple book keeping for energy and momentum for distributed objects like chains.
A length s of chain has mass s. If the speed is v, then the momentum is sv and the kinetic energy
is 12 sv2 . If a chain is being dragged out of a pile of stationary chain with speed v, then the length s already out of the pile and in motion increases at a rate ds/ dt = v. Thus the rate of momentum change is
d(sv)/ dt = v ds/ dt = v2 . Likewise the rate of change of kinetic energy is 12 v3 (derive this!).
Exercise 3: Picking up a chain at constant speed, v, neglecting gravity. Show that a force v2 must be exerted
on the end of a chain to start setting it in motion. What power is exerted by the force? Compare this with
the rate of creation of kinetic energy. Comment!
Hint: Force is the rate of change of momentum. [MW]
Exercise 4: Consider a long length of chain in a pile on a flat surface. One end is raised against gravity g at
a constant speed v by applying an upward force f = T (0), the tension at the end of the chain. Show that
when a length L of chain has been raised, then T (0) = Lg + v2 . Further determine the power input by f
and compare this with the rate of change of stored energy in the chain. [MW]

Discussion of energy and momentum in distributed objects

The 12 in KE = 12 v2 is not consistent with the momentum flux3 expression v2 . One half of the energy is
lost during start up of the motion by any force that is doing work while providing momentum. It is the
same in many such problems in mechanics, for instance when ore or coal is put on a moving conveyor
belt. We also see it in other branches of physics, for instance connecting a capacitance C across a battery
delivering charge at a potential V takes a charge Q = CV from the battery at potential V. Thus the battery
delivers work QV = CV 2 . But the energy stored in the capacitor when it fully charged is only 12 CV 2 .
The problem is that charge is delivered to the capacitor through a finite potential difference from V to
the current potential of the capacitor. (The difference gets ever smaller as the capacitor charges up.) That
portion of the energy is not storable, something that is avoided in AC circuits where the potential varies
with time and matches that across the capacitor as charge flows in.

The motion of chains and ropes around corners

The motion of ropes along their own length is most surprising. It takes us a step closer to the chain fountain
problem. It was much studied in the 1850s (including by the Astronomer Royal) since there was industrial
interest in laying submarine cables for telecommunications. The problems found their way into the Cambridge exams certainly by 1854 and then by 1860 into main-stream mechanics text books, for instance that
of Routh.
Consider a rope of mass/unit length under tension T going around a pulley at a speed v; see figure 4.
Let us show that if the speed and tension are appropriately related, then one does not need the pulley at all!
[Ignore gravity for the moment.] Results for centripetal motion are required [see the RSPP concept page on
circular motion, and the relevant questions] and we sketch them here.
In circular motion, the inward radial acceleration is v2 /r, with r the circles (here, the pulleys) radius;
see figure 5. A length of arc ds has mass ds and hence a force (v2 /r) ds is required (from f = ma) to
accelerate the element inwards as it goes around the arc. The effects of tensions in a circular arc of rope are
similar to the effects of a tangential v leading to v2 /r. The two forces T (s) and T = T (s + ds) in figure 5
are equal in magnitude but not quite opposite in direction. Their difference dT = T (s + ds) T (s) points
inward, see the vector triangle of figure 5 and is of magnitude T d; think of dT as a small arc of a circle of
radius T. But from the actual circle ds = r d, or d = ds/r whereupon dT = T ds/r. The inward force per
unit length (dividing by ds) is T/r.
3 See

Flux in the Maths concepts pages on vectors.

v
v

T
d

Figure 4: A rope under tension T

travelling around a pulley at speed
v.

-v

dv

T
r
d

dT

-T

Figure 5: Centripetal acceleration and inward forces from tangential motion

and tangential tensions.

This force generates the centripetal acceleration:

v2 /r = T/r v2 = T.
The strange consequence of 1/r cancelling on both sides is that, if T = v2 , then a rope moving along
its own length can turn any corner (r has disappeared from our considerations). We could dispose of the
pulley altogether; indeed any other curve in the rope is stable.

A research problem the chain fountain analysed

** You now have the machinery to analyse a difficult problem with the astonishing result seen in figure 1.
This section can be skipped at a first reading if you want to proceed to catenaries, bridges, arches etc.
Figure 6 shows a schematic of the fountain that serves as a first model to analyse
the fountain. The chain in a pile on the table rises up a height h2 , whereupon it
turns around and then drops a height h2 + h1 to the floor, so the net drop is h1 .
Tensions are marked at several points in the chain. In the curved region we know
that TC = v2 . In saying this we neglect the effect of gravity in the curved region
in adding to the centripetal acceleration. It amounts to an assumption that r h2
and r h1 . To account for gravity at the very tip of the arch involves a catenary
analysis, which is done below, and the shape is not semi-circular. The results are
qualitatively the same. The tension at the floor is marked TF at the floor. A first
assumption is that T = 0 there with no chain below there can be no weight
putting the chain under tension, which for most chains is reasonable and one can
check this is the case for Moulds chain. It is not true for all chains and in effect a
smooth, inert table can pull down with TF , which is a shocking result presented
by Grewal et al; see http://www.youtube.com/watch?v=i9gLi4pBgpk.

TC
v

h2

TT
Table

v
h1
TF

Floor

Figure 6: A model for the

chain fountain of figure 1.

The chain fountain 1 naively, it does not exist!

The most obvious value for the tension at the table, TT , is that required to accelerate the chain to speed v, that is TT = v2 which we proved is the force required
to provide the momentum of the chain. We also have the relations between the
tensions at different parts of the chain due to the dead weight of the sections between them. In summary:
TC = v2

TC = TT + h2 g

TC = TF + (h2 + h1 ) g

where we will take TF = 0. One can immediately see that the first two of these equations are inconsistent
with each other if TT = v2 unless also h2 = 0. If the tension in the chain at the point of pickup is providing
all the momentum to set the chain in motion, then there can be no chain fountain: h2 = 0 means the chain
just falls over the edge of the table without rising.

The chain fountain 2 it does exist, but something additional is pushing the chain at lift off!
Since Mould observed a fountain, then there must be a source of momentum additional to TT that is accelerating the chain. This can only be from the table pushing upwards with a force R = v2 . Here is
a constant (less than a 21 as it turns out) and the rest of R is of the form v2 on dimensional grounds since
the only function of such a force, were it to exist, is to accelerate the chain (the momentum flow of which
is v2 ). That there is an upward force from the chain pile on the departing chain, is also shocking and is
discussed below. For the moment one has to accept from momentum balance and the fact that fountains
exist, there must be such a force. Thus the total force accelerating the chain breaks down as:
Ttotal = v2 = TT + v2
whereupon TT = (1 )v2 .
Exercise 5: Consider momentum initially imparted to the chain, and tension variation along the chain from
the pick up point to the apex and then to the landing point. Show that the rise and speed of the chain is:
v2 = h1 g/(1 ).

h2 = h1 /(1 ),

Show also that the ratio of kinetic energy to potential energy change between the table and the floor is
KE/PE = 12 v2 /( gh1 ) =

1
.
2( 1 )

Thus place bounds on and find the height of the highest fountain in terms of the drop h1 . Note that
observed fountain heights for the Mould-type chains are h2 0.14h1 . [MW]

Why does the table push the chain at lift off?

Consider the physical origin of the reaction force R = v2 needed for the chain fountain. Model the chain
as a set of freely jointed rods each with length b, mass m and moment of inertia I. Links being set into
motion are of necessity mostly at rest with a horizontal orientation, and are pulled at one end by preceding
links that are directed largely upward. See figure 7 where a single link is picked up by a vertical force TT
applied at one of its ends. This upward force induces it to both rise and rotate. If the rod were in free space,
the other end of the link would move down. When the rod sits on a horizontal surface (being the table or
the rest of the pile of chain) then the pile supplies an additional upwards reaction force R to prevent this
end of the link moving down. We pose a problem that gives an estimate of R in the initial phase of lift-up
which we take as representative of R overall. You might like to consult the RSPP pages and problems on
angular motion before attempting this question.
Exercise 6: Solve for the initial stages of the links motion where there is a linear acceleration a and an
We use the initial force values as estimates of those acting during pickup.
angular acceleration .

Show that to avoid the end at R being rotated through the table, one needs a = (b/2).







 



Figure 7: Left: A rigid rod of mass m and moment of inertia I lies on a horizontal surface (in practice the pile of
chain) and is picked up via a vertical force TT applied at one end causing the rods center of mass to rise at a speed
v. In order for the rod not to penetrate the surface, the surface must also provide a vertical reaction force R on the
opposite end of the rod. Middle and Right: The ball chain in our experiments required 6 beads to turn by (middle)
so we model a link of the chain as consisting of 3 identical point masses (beads) connected by massless rods (right).

An R acts to achieve consistent rates of change of linear and angular momentum required to prevent the
R-end rotating through the table. Show that
TT + R = ma

and

( TT R)(b/2) = I .

From these equations and using = a/(b/2), show that

I
1
R = ma 1 1
2
2
4 mb

Relating ma to the forces and hence the upward momentum flow on pick up, show that
!
I
1
R
1 1
=
.
=
2
v2
2
mb
4
This expression gives an upper limit on that conserves energy; see the previous exercise where this issue
is confronted. Estimate I of an effective link by recognizing that large bend in the chain is only achievable
after traversing at least a few beads, figure 7. The number of beads required is an estimate of the length the
equivalent rigid rod making up the chain. Find values of and connect with observed h2 /h1 values. [MW]
This analysis completes a simple analysis of the chain fountain. We now discuss the shape of static
hanging chains, their relation to bridges and, in inverted form, to arches. One could then return to a
detailed analysis of the chain fountain, which is an inverted catenary.

The shape of a hanging chain 2

We derive curve formed by the hanging chain, shown in figure 3.
Exercise 7: Recall from Ex. 2 that tan (s) =

s
a

is the intrinsic form of a catenary curve for a hanging chain.

Prove that with Cartesian coordinates x and y, along with the arc length s measured from the lowest point, one has
q
s/a = sinh( x/a)
y( x)/a = cosh( x/a)
y(s)/a = 1 + (s/a)2 ,

where the central y( x) form is the Cartesian equation for a catenary.

q
Hint: Prove that a small length along the chain s =
(x)2 + (y)2 =
r
 2
y
x 1 + x . Explore the connection between this relation and (s). The trian-

(s) s

x
Figure 8: A small enough
section of the curve y( x )
is approximately straight.
Apply Pythagoras to the
small triangle.

gle in Figure 8 shows a small section of the catenary. First find the the connection
between s and x. [MW & PS.]

A summary of forms and relations for a catenary; what it all means

The catenary has a shape given either as y( x) or (s), or combinations of these variables. Also, the tension
varies along the chain. We summarise the relations below. Test your understanding by reproducing and
converting between them.
p
s/a = sinh( x/a)
y/a = 1 + (s/a)2
y( x) = a cosh( x/a)
tan (s) = s/a

T (s) = T (0) sec (s)

q
T (s) = T (0) 1 + (s/a)2

y/a = sec (s)

T (y) = T (0)y/a
T (y) = gy

a = T (0)/g

T ( x) = T (0) cosh( x/a)

y( x ) a

Explore the limits of these expressions and the role of the total length of chain, the separation of the end
points, etc. Note that the angle (s) only approaches the vertical ( /2) as L/a 1 (locate the
appropriate connection above) where s = L is one end of the chain of total length 2L. If the x coordinate of
right chain end is x L and it approaches L, the chain is nearly stretched out tight. In that event, ( L) 0
and the above relations give a L, both in the connection between and s and in the connection between
s and x (check both). The tension is then enormous the chain is very flat, supporting its weight with
only a small component of its tension being in the vertical direction. See this from T (0) = ag Lg, the
latter being the simple deadweight of the chain. Check the T (s) relation then shows little variation with s.

A suspension bridge
Very strong, hanging cables support bridge carriageways by suspending cables between the strong major
cable and the road. Now the mass per unit length is not from per unit length of the cable, but from per
unit length of the load, i.e. the horizontal carriageway. This simplifies things greatly: the coordinate upon
which angle and mass depends is simply x, the Cartesian coordinate, so = ( x) etc.
x
major cable

(a)

y(x)

pier

mass/area
(rubble)
arch

(b)

carriageway, mass per unit length

Figure 9: (a) A suspension bridge where the major cable supports a carriageway. (b) A Roman bridge in the form of
a light arch supporting a rubble fill above it on which a carriageway sits.

Exercise 8: A suspension bridge is held up by light cables connected to a major cable supporting a carriageway of mass per unit length. It is supported by a pier at either end (see Figure 9(a)). Show that the major
2
cable takes the shape y = 12 xa for suitable origins of x and y, and find an expression for a in terms of , g
and the tension in the lowest part of the chain.
Hint: Consider the static equilibrium of a section of chain from x = 0 to x. [MW & CL]

Arches; a Roman bridge

Recall that chain, ropes, threads etc. can only be in tension (no compression or shear/deflection). Likewise
masonry can only be in compression. So masonry arches are like chains, with the signs of force & gravity
reversed, and are analogies of the hanging structures we have examined above. So a pure arch, supporting
nothing other than itself, would be an inverted catenary (prove this!). Arches that support things will need
to be a different shape, just as we have seen that hanging ropes that support a bridge are not catenaries:
Exercise 9: A Roman method of bridge building was to form a masonry arch and fill in above it with rubble;
figure 9(b). The arch is only stable if the lines of compressive stress follow the path of the arch. If the mass
per
R x unit area above the arch is , show that the mass of rubble above the arch between x = 0 and x is
0 y( x ) dx . By Rconsidering the equilibrium of the light arch supporting this section of bridge, show that
x
g
tan(( x)) = T (0) 0 y( x )dx where T (0) is the compressive tension in the highest part of the arch, and ( x)
is the angle between the tangent to the arch at x and the x axis. Show further that the shape required for the
supporting arch must be of the form y = y0 cosh ( x/a) . Give an expression for a, and discuss the meaning
of y0 . [MW & CL]

[MW & JSB with help from JC-Z & PS; Oct. 2013]